سیاست های پولی در آلمان : یک تجزیه و تحلیل هم انباشتگی در رابطه با قوانین نرخ بهره

The paper attempts to identify an empirical relationship that characterizes the way the Bundesbank adjusted its short-term rate with respect to various objectives. By building on a careful exploration of the properties of the variables involved, it is established that interest rate rules —often remarkably similar to the Taylor rule— remain valid and relevant in a Vector Error Correction framework, and thereby proposing a distinctive interpretation of German monetary policy during the period 1975–1998.

The German Central Bank, the Deutsche Bundesbank (DBB henceforth), is commonly associated with the concept of monetary targeting. However, operationally, its policy involves the setting of the short-term interest rate, or, in other words, the translation of its main goals into interest rate objectives. The present paper is based on a careful exploration of the properties of the related variables and attempts to identify an empirical relationship that characterizes the way that the Bundesbank adjusted its short-term rate over time. The estimation of this relationship reveals the implicit way the bank's decisions translated into a reaction function and is of interest since very often EMU monetary policy is compared to what it is believed the Bundesbank would have done.1 Although the DBB is no longer responsible for policy setting, an analysis of the German experience is indeed most relevant for at least the following reasons. First, Germany is a major economy of the EMU and as such matters for decision making by the European Central Bank (ECB henceforth): the subscription of the DBB to the capital of the ECB is the highest among all the banks of the European System of Central Banks. Second, the ECB operates within a framework very similar to the one of the DBB, in an attempt to inherit good reputation and to cope with the uncertainties of the starting period. Third, the DBB used to be a leading monetary authority (both internationally and within the European Monetary System) that followed an independent monetary policy. Its performance is judged, by international standards, as strikingly good given that the level and the fluctuations of the domestic inflation rate over time were among the lowest. Such a stable inflation environment is likely to be representative for the euro area, as well as for other economies worldwide. Therefore, all in all, analysing policy setting by the Bundesbank provides significant insights for the conduct of monetary policy. There are numerous empirical studies on monetary policy in Germany. One strand of the literature focuses on the modelling of a money demand relation in various frameworks —for a brief review see Lütkepohl and Wolters (1998). Another strand of the literature acknowledges in the German monetary targeting regime key elements of inflation targeting. In this context, Mishkin and Posen (1997) have offered a narrative comparative study and, as regards some often cited relevant empirical studies, Bernanke and Mihov (1997) have noted that forecasted inflation explained a greater share of the variance of German monetary policy than did forecasted money growth, while Clarida and Gertler (1997) and Clarida et al. (1998) have shown, by means of distinct approaches, that the DBB was adjusting the short-term interest rates according to an interest rate rule. However, crucial properties of the data, like for instance the integration properties, are often ignored in this empirical literature.2 Within the context of interest rate rules, there is a growing literature concerned with these shortcomings. The root of the matter is the finding of Granger and Newbold (1974) and Phillips, 1986 and Phillips, 1989 that if variables integrated of order one are found not to be cointegrated, a static regression in levels is spurious. As regards up-to-date research, Bunzel and Enders (2005) and Siklos and Wohar (2006) have discussed a number of issues, have reported evidence of non-stationarity of the involved variables (with US data) and, as Christensen and Nielsen (2003), have experimented with long-run cointegrating relationships. Furthermore, Gerlach-Kristen (2003) and Österholm (2005) have explored the econometric properties of the Taylor rule and have found signs of instability, misspecification and inconsistencies due to the mistreatment of the non-stationarity of the data. The present paper begins with a thorough analysis of the features of the data generating process (DGP) and, in the context of the literature on inflation targeting, elaborates on an interest rate relationship, namely the Taylor (1993) rule. This is a so-called ‘simple’ rule originally designed to track policy setting in the United States that has become a rather popular benchmark: it calculates an economy's best interest rate value as a function of its state, which is described by the deviation of actual inflation rate from a target and of actual output from its long-run potential.3 The analysis is performed by means of a trivariate vector error correction model (VEC model henceforth), which comprises an output variable and inflation apart from the short-term interest rate. A model that includes a measure of the money stock is without any doubt required, given the privileged role attributed to money growth in the DBB's announcements; besides, exploring the implications of such a model may be a way to deal with the point raised by Minford et al. (2002), who have argued that the empirically estimated Taylor-type rule reaction function can be a reduced form of a monetary policy rule with the money supply hidden in the residual. By adding variables like the US overnight rate or a long-term rate, I investigate whether the DBB responded to other indicators.4 The analysis examines a complete historical period of German monetary policy, as it covers the period from roughly 1975 to 1998. In 1975, shortly after the breakdown of the Bretton Woods system of fixed exchange rates, the first annual monetary target was announced by the DBB. According to its statute, the bank was bound to ‘safeguard the currency’, which, in Issing (1997), was interpreted to mean price stability. By means of a procedure that remained in principle unchanged, the bank was setting targets for monetary growth that implicitly incorporated goals for inflation-for a comprehensive presentation of the German monetary policy see Neumann and von Hagen (1993), Baltensperger (1999) or Schmid (1999). On 1 January, 1999 the responsibility for the conduct of monetary policy was handed to the ECB. The paper is organized as follows: in the next section, I provide explanatory information on the dataseries and the econometric framework utilized; in the third section, I present the results of the empirical analysis for each model and, in the last section, I offer some concluding remarks. In a nutshell, I demonstrate that a stable interest rate rule, very similar to the popular Taylor (1993) rule, emerges repeatedly as the long-term relationship connecting the policy rate with output and inflation. Thereby, an untraditional standpoint on the DBB's monetary policy is put forward and an alternative methodological approach is established.

The paper establishes a framework for analysing monetary policy setting in a key economy of the EMU and, since this economy experienced an admirable performance as regards price stability, the findings of the paper are of use to policymakers seeking a similar objective. To be more specific, the study explores the way the short-term rate of the Bundesbank adjusted to various objectives during the period 1975 to 1998. As demonstrated, the common stochastic trend that emerges among the policy rate, the domestic inflation rate and the measure of economic activity qualifies as an interest rate rule, which is comparable to the Taylor (1993) rule as far as the derived parameter estimates are concerned. Interestingly, this long-run interest rate rule, expressed via robust parameter estimates and dynamics emerges repeatedly from models featuring various pertinent variables. What is more, in general these models generate reasonable dynamic behavior. The higher dimension models explain how the policy rate of the Bundesbank reacted to movements in other indicators. More specifically, the inclusion of the Federal Funds rate as a means of capturing the importance of foreign constraints in the adjustment of the German policy instrument leaves the reactions to the inflation rate and output virtually unaffected, and generates a positive impact on the policy rate. Fascinatingly, the inclusion of the real money stock does not disturb the size of the Taylor-rule parameter estimates. Most importantly, in all cases, and independently of the composition of the model, the impulse response analysis illustrates that the response of the day-to-day rate to shocks in output and inflation is perceptibly robust and consistent with the predictions of a Taylor-type interest rate rule. Furthermore, it is demonstrated that the Taylor-type interest rate rule remains relevant in a representation where the stochastic properties of the variables are adequately modelled, outwitting a thorny criticism often encountered in the literature. In this way, the paper contributes to the current empirical literature by leading the way in employing an alternative framework with interesting qualities for analyzing monetary policy. Given that the rule is incorporated into an error correction model, some of its usual features are not present in the standard way, but instead are captured indirectly through the short-term dynamic structure of the model. In particular, despite the absence of the interest rate smoothing parameter, the partial adjustment is captured by the dynamics of the model. Likewise, although there is no constant term as such in the cointegrating relationship —the reason for this being the inclusion of the trend that encapsulates it in some way—, it is included in the deterministics of the model. Moreover, the output gap is not exogenously measured, but is captured within the model as the deviation of the real output from its trend. To conclude, the study can be extended to countries which have officially adopted some form of inflation targeting as a framework for monetary policy-making. Such an experiment will allow us to explore how the functioning and the dynamics of the model are affected.